Protected areas aim to conserve nature, ecosystem services, and cultural values; however, they have variable success in doing so under high development pressure. Southeast Asian protected areas faced the highest level of human pressure at the turn of the twenty-first century. To estimate their effectiveness in conserving forest cover and forest carbon stocks for 2000–2018, we used statistical matching methods to control for the non-random location of protected areas, to compare protection against a matched counterfactual. We found Southeast Asian protected areas had three times less forest cover loss than similar landscapes without protection. Protected areas that had completed management reporting using the Management Effectiveness Tracking Tool (METT) conserved significantly more forest cover and forest carbon stocks than those that had not. Management scores were positively associated with the level of carbon emissions avoided, but not the level of forest cover loss avoided. Our study is the first to find that METT scores could predict the level of carbon emissions avoided in protected areas. Given that only 11% of protected areas in Southeast Asia had completed METT surveys, our results illustrate the need to scale-up protected area management effectiveness reporting programs to improve their effectiveness for conserving forests, and for storing and sequestering carbon.
Natural forests have high conservation value in terms of irreplaceable habitat for a disproportionately large number of species as well as for the ecosystem services they provide, such as carbon sequestration and storage1,2. Populations of vertebrates have declined by 68% on average between 1970 and 20163 and the extent of ‘healthy’ forests has shrunk to 40% of pre-human extent4. Some scientists warn that forests have gone beyond the precautionary ‘safe limit’ for land-system change as the level of forest cover loss has disturbed the biogeophysical processes of earth systems that directly regulate climate5. Protecting forests from intense levels of human pressure is essential for stemming the accelerating loss of species and their habitats, and for maintaining ecosystem services6. The last three decades have seen a considerable expansion of terrestrial and marine protected areas, which coincides with targets agreed under the Convention on Biological Diversity, effective from 1993. However, expanding the coverage of protected areas alone has proved insufficient to protect biodiversity; effective management within protected area borders is an essential component of ensuring species’ populations are stable or improving7.
Rigorous evaluations of protected area performance remain relatively sparse, especially in the tropical zone8. Evidence depicting why some protected areas perform well, while others do not, is even scarcer9, yet this knowledge is crucial to improve the effectiveness of the protected area network. Further, many previous studies fail to account for the non-random location of protected areas by comparing areas that are protected to all the unprotected areas present in the landscape10, or to neighbouring buffers11. These approaches ignore any bias in protected area placement towards locations that are remote (i.e., high altitude, steep, far from urban centres or roads) and have less potential for agricultural production, which in turn affects the level of deforestation pressure11,12,13. Changes in forest cover or human pressure are two widely used indicators to measure the level of exposure to threats14,15,16. The advantage of such indicators is their broad coverage, that increases the comparability across sites. Separating the impact of law enforcement and management from protected area location can be achieved by comparing protected areas to counterfactual sites with similar characteristics. When using a counterfactual approach, estimates of the impact of protection on avoiding deforestation were consistently reduced17,18,19.
Based on Landsat satellite imagery between 2000–2013, it is estimated the global protected area network lost 3.4 times less intact forest than unprotected areas, accounting for differences in accessibility, terrain and human pressure20. Geographically, protection had the weakest effect on maintaining intact forests in Southeast Asia, Australia and tropical South America20. Indo-Malayan forests have exceptional biological diversity, yet also rapid rates of deforestation21, driven by global demand for timber, palm oil and rubber22. Protected areas in Southeast Asia experienced a higher mean change in human pressure between 1995 and 2010 than in any other region in the world14. Given historical deforestation rates, Sodhi et al.23 warned that over 40% of the region's biodiversity may vanish by 2100. Understanding the impact of interventions in this region, like protection status or carbon payments on retaining forest cover, is critical for improving conservation responses with finite resources and is a requirement for satisfying Monitoring, Evaluation and Reporting (MER) program requirements to access REDD+ and other funding streams (e.g., the Global Environmental Facility and the World Bank). Stronger forest protection and conservation efforts are needed across Southeast Asia’s existing protected areas to avert the future trajectories of forest cover and forest carbon loss estimated by 205024.
Here, we evaluate the effectiveness of Southeast Asian protected areas at reducing forest cover loss and associated carbon loss, while controlling for spatial variation in deforestation pressure. We also explore whether the process of conducting management effectiveness reporting is correlated to improved protected area performance, and whether performance can be predicted by levels of management resourcing. Protection from logging, expansion of monoculture plantations and other threats is dependent on site-level factors, such as adequate investment in management activities, as well as system-level factors, such as national governance, yet the level of protection afforded by it is complex. We wanted to understand how these factors inter-relate, and link to the level of avoided forest cover loss and carbon emissions. Our work has implications for the usefulness of the Management Effectiveness Tracking Tool (METT)25 (a score-card to assess the adequacy of management in protected areas based on the IUCN management effectiveness framework and used in over 3,000 sites across the globe)26, for predicting conservation impact and for two global environmental agreements. Parties to the Convention on Biological Diversity27 are now preparing and revising their post-2020 biodiversity decadal targets and parties to the UN Framework Convention on Climate Change28 revise their nationally determined contributions to reduce emissions under the Paris Climate Agreement every five years. Both conventions cover commitments to reduce deforestation and ecosystem service loss through the long-term protection of natural systems, and it is crucial to understand the contribution of protected areas towards achieving these targets.
Effectiveness of protected areas for reducing forest cover loss and carbon emissions
Our results show that protected areas significantly reduced the loss of forest cover with the overall rate of forest cover loss across the region being three times lower inside protected areas (n = 692) over the period 2000 to 2018 than in the matched unprotected landscape (p < 0.001; df = 76,679, t = − 49.55; Table 1). However, this overall effect masked important differences between countries, with Malaysia having the most effective protected area network, avoiding 14.57% of forest cover loss, followed by Cambodia (11.16%), Vietnam (10.27%), Laos (10.31%), Thailand (5.62%), Indonesia (3.93%), and Myanmar (2.28%). In the Philippines, protected areas lost 3 times more forest cover than unprotected areas, meaning not only was protection not effective, it in fact accelerated forest loss. Likewise, each country-level effect concealed differences between protected areas (Fig. 1).
Carbon emissions were likewise significantly lower in protected areas over the period 2000 to 2018 than in matched areas not under protection (p < 0.001; df = 76,679, t = − 67.87). Post-matching, the average rate of carbon emissions in protected areas was 2.5 times lower (1695t CO2 per km2) than the counterfactual (4313t CO2 per km2), resulting in a net treatment effect of 2618t CO2 per km2, equivalent to a saving of 145t CO2 per km2 per year.
The difference between the naïve (pre-matching) and counterfactual (post-matching) estimates of avoided forest cover loss and carbon emissions was small on average for the region (Table 1). However, for some countries, there were large changes in the magnitude of avoided forest cover and carbon emissions between pre- and post-matching and the direction of these changes varied inconsistently (Fig. 2). Estimates of avoided forest cover loss and carbon emissions decreased in 3 countries, increased in 3 countries and remained consistent in 2 countries. For example, Indonesian protected areas decreased from being 5 times more effective at reducing forest cover loss before matching, to only twice as effective than the counterfactual post-matching. Vietnam’s protected areas were 12 times more effective than the counterfactual prior to matching, yet after matching this factor increased to 16. Post-matching estimates of avoided forest cover loss and carbon emissions increased for Thailand, though the matching performance was poor, meaning the matching did not yield a large improvement in balancing the treatment and the counterfactual samples (S1).
Impact of protected area monitoring and management reporting
Our results show that protected areas with management effectiveness assessments had significantly reduced rates of forest cover loss and carbon emissions, with the overall rate of forest cover loss (p = 0.004) and carbon emissions (p = 0.016) significantly lower in protected areas that had completed METT assessments, compared to those that had not (Figs. 3 and 4). The average rate of forest cover loss avoided in METT protected areas was 9.67% (n = 73), compared to 8.10% in non-METT protected areas (n = 619). The average rate of carbon emissions avoided in protected areas that had METT assessments was 3500t CO2 per km2, compared to 2942t CO2 per km2 in the non-METT protected areas.
When exploring the individual components of the METT in more detail (Fig. 5), our most parsimonious model for predicting avoided forest cover loss did not detect a relationship between any of the management dimensions or contextual factors and avoided forest cover loss. However, for avoided carbon emissions, our most parsimonious model included protected area level management resourcing, design and planning, governance, national-level transparency, years protected, protected area size, human footprint index, oil palm suitability, and slope (Fig. 5). Country as a random effect explained 18% of the variance in the data.
Despite the intense level of human pressure Southeast Asian forests are facing14, we found the protected area network in the region lost 3 times less forest cover between 2000–2018 than the analogous unprotected landscape. This amounted to 2.5 times less carbon emissions in protected areas than the counterfactual. Our estimate is consistent with the findings from a global study20 that estimated Southeast Asian protected areas lost 3.3 times less intact forest between 2000–2013 than unprotected areas. These results also add to previous counterfactual matching studies from Sumatra and Borneo, that show protected areas resisted deforestation pressure over earlier periods18,29.
We found that statistical matching, which controlled for the non-random location of protected areas, had a variable effect on the direction and magnitude of our estimates of the protection impact. We found that post-matching estimates of avoided forest cover loss and carbon emissions decreased for Indonesia, Malaysia and Myanmar, remained stable for Laos and the Philippines, and increased for Thailand, Vietnam and Cambodia. Our finding contrasts with the consensus from the literature that the matched counterfactual approach consistently reduces the estimated impact of protection17,18,19. Joppa and Pfaff (2010) explain the reduced impact post-matching by the general bias in protected area placement towards places subject to low human pressure, which overestimates the protection effect. We did not find this pattern consistently across all countries. Bebber and Butt30 observed that controlling for residual placement bias was more important for regions with high forest cover. We suggest the causation mechanism underlying this pattern is for countries with high remaining forest cover, deforestation activities will be targeted towards unprotected forests over protected forests, whereas for countries with low remaining forest cover, protected areas must withstand escalating deforestation pressure due to lack of choice in available forest resources31. This theory offers a plausible explanation for Vietnam and the Philippines. Protected areas in the Philippines have the highest level of human pressure, yet protection appears to have the perverse impact of accelerating forest and forest carbon loss; a trend particularly evident on the island of Palawan where protected areas experienced more forest loss than expected32. We do not attempt to infer patterns of causation for the matched results for Thailand and Cambodia because the matching performance was poor (S1). To synthesize, we infer that these patterns may be driven by human pressure and proportion of remaining forest cover, though we recommend further in-depth exploration of the local socioeconomic conditions that may be driving this trend.
Our findings show that protection resulted in statistically significant lower levels of forest cover loss and carbon emissions, compared to the unprotected analogous landscape. Protected areas in the region have on average withstood escalating pressure relative to the analogous unprotected landscape in Cambodia, Indonesia, Laos, Malaysia, Myanmar, Thailand and Vietnam, but not in the Philippines. Regional deforestation trends in Southeast Asia are influenced largely by Indonesia, where over 50% of the region’s forests remain and where ~ 40% of protected areas are located. In 2017, the annual deforestation rate in Indonesia dropped by 60%, followed by another small decline in 2018. These positive results may be influenced by policy initiatives to slow deforestation and restore peatlands in Indonesia, such as Presidential Instruction No. 10/2011 suspending new concession licenses for logging and forest conversion in primary forest and peatlands, and the high volume of REDD+ projects33,34. Despite overall positive outcomes of protection, there are examples where protected areas were completely ineffective. For example, two Cambodian protected areas Roneam Daun Sam Wildlife Sanctuary and Snoul Wildlife Sanctuary lost 100% and 80% of forest cover respectively and both were degazetted by the Government of Cambodia shortly after. These examples highlight that without adequate resourcing, political backing or community support, protected areas can be degraded and degazetted for increased access to extractive activities35.
Protected areas with Management Effectiveness Tracking Tool (METT) assessments withstood significantly more forest cover loss and carbon emissions than non-METT protected areas. This finding highlights the value of investing in monitoring and evaluation programs to predict performance. We recommend further exploration to determine the causal link in this relationship. By benchmarking performance against best practice standards, protected area managers may be identifying key strengths and weaknesses and thereby improving future performance as a result of the monitoring process. We also note that as METT assessments are a requirement of GEF-funded projects, the protected areas attracting donor funding may be funded because they have better performance. Given our findings that management scores are positively linked to carbon savings, combined with growing evidence of positive correlations between management scores and biodiversity population trends36,37,38, the value gained by participating in protected area management effectiveness reporting programs is becoming clearer39. Yet only 11% of protected areas in Southeast Asia had completed METT surveys.
Our study is the first to find evidence that management effectiveness scores from the METT database can predict the level of carbon emissions avoided in protected areas. All other studies to our knowledge, that attempted to link avoided deforestation or reduced fire incidents to management scores at sites across South America and Madagascar, did not detect a clear pattern40,41,42,43,45. Further, most of these studies attempt to link management scores to deforestation or fire, not carbon emissions. Although we found avoided carbon emissions and forest cover loss were correlated (Supplementary Information), some key differences arose. Differences can be due to woody vegetation density, vegetation type, stem diameter at breast height and vegetation age46. This is particularly relevant given our increased knowledge of the contribution of tropical protected areas in achieving targets laid out in the Paris Agreement30,47. Tropical protected areas reduced ~ 29% of tropical deforestation emissions between 2000–2012, compared to expected deforestation30. Looking forward, the conservation, restoration, and improved management of tropical forests, mangroves, and peatlands could provide between 23–37% of the cost-effective mitigation solution required by 2030 to limit global warming to 2 °C48,49. To account for losses in natural forests that are replaced by monoculture plantations (e.g., rubber, oil palm), we recommend future research explore carbon trends to provide deeper insight into forest cover change dynamics.
One of the most important caveats to consider when interpreting these results is that we have not considered land tenure and land use categories in matched areas not under formal protection, as described in Schleicher et al.50. Land that aligns with the IUCN definition of a protected area and contained in the World Database of Protected Areas (WDPA), does not capture all land with conservation value. Community-managed forests are a growing governance arrangement in developing countries51 and have reduced rates of deforestation in Indonesia52. Separating the different types of land use in the counterfactual area not under protection would provide further insight into what is occurring outside the protected area estate. However, other effective area-based conservation measures (OECMs) are not listed in the WDPA, unless they are declared. Similarly, we grouped different types of protected areas into a singular treatment group, when in reality, there are varying levels of human perturbations allowed under different governance regimes53. Other studies have found that protected area strictness does not consistently improve environmental and social outcomes12,54. Future research assessing how protected area management and governance dimensions relate to social impacts would be highly valuable to understand their multi-dimensional impacts. Evidence exists of positive conservation and socioeconomic outcomes from protected areas55, particularly those that implement collaborative management regimes, where local people were empowered and maintained cultural and livelihood benefits56; yet no studies to our knowledge have linked management scores with social impacts.
In conclusion, by accounting for spatially-dynamic deforestation pressures, we find evidence that the Southeast Asian protected area network is conserving forests and forest carbon stocks, in addition to the biodiversity benefits previously reported37. Higher levels of management resourcing are associated with greater reductions in carbon emissions. The protected area network in the region therefore presents strong opportunities for scaled-up investment of conservation and REDD + finance. Stronger forest protection and conservation efforts are needed in Southeast Asia’s existing protected areas to avert projected trajectories of forest cover and forest carbon loss estimated by 205024. However, protection requires funding to implement and less than 3% of all climate mitigation funding went to forest-based mitigation strategies57,58 One step towards mobilizing greater financial support is demonstrating strong links between management resourcing and conservation impacts. Given the low level of finance being directed at forest-based climate mitigation solutions, combined with the sizable contribution of tropical protected areas in achieving the Paris Agreement targets30,48,59, we recommend that financiers enhance stimulus for protection to more effectively store and sequester carbon. Our study shows that management effectiveness frameworks, such as the METT, can play a guiding role in predicting protection outcomes.
Materials and methods
We analysed a total area of 747,714 protected and unprotected pixels, covering eight countries: Cambodia, Indonesia, Laos, Malaysia, Myanmar, the Philippines, Thailand, and Vietnam. Timor-Leste, Brunei, and Singapore were excluded from the analysis because there were insufficient pixels in these countries to do matching. Full details of our data sources and preparation are provided in Supplementary Information with a summary included below. All the data we used is freely available. All spatial analysis was performed in ArcMap v10.5 (ESRI 2016) using Asia South Albers Equal Area Conic projection. Statistical modeling was performed in R v3.4.3 (R Development Core Team 2017).
Avoided forest cover loss and carbon emissions
We used the Hansen tree cover loss maps that measure the annual change in tree cover between 2000 and 2018 and the tree canopy cover map for the year 2000 at a spatial resolution of ∼30 m across the terrestrial world (version 1.6; Hansen et al.60). We used the aggregate tool in ArcMap to downscale from the ~ 30 m pixels to ~ 1 km resolution (Supplementary Information). Tree cover loss includes either outright deforestation or temporary disturbance (e.g. due to forestry, selective logging, shifting cultivation or wildfires61). To estimate forest cover loss between 2000 and 2018, we created a binary forest cover map for the year 2000, where forest was categorized as having tree cover of greater than or equal to 40% canopy cover and “loss” was restricted to only those pixels that were classified as “forest” in 2000. We identified this threshold by running comparisons of the total forest area resulting from a range of thresholds and comparing it against the total forest area in the Global Forest Watch platform (www.globalforestwatch.org). The rate of forest cover loss in each protected area was calculated by aggregating the total number of pixels that had lost forest cover over the 18-year period and dividing this total by the number of pixels sampled. For carbon emissions, we used the CO2 emissions from aboveground woody biomass loss between 2000 and 2018 at a resolution of ∼30 m across the tropics62. In this study, “carbon emissions” refers to emissions “committed” at the time of disturbance or clearing, noting there may be a time lag until they are “realized”. Again, we used the aggregate tool in ArcMap to downscale from the ~ 30 m pixels to ~ 1 km resolution by summing all values. Carbon emissions per protected area were calculated in the same way as forest cover loss, resulting in an average carbon emission saving per pixel. Our carbon emission estimates do not include emissions from peat.
We measured the treatment effect in terms of deforestation for each country, by dividing the average forest cover loss rate in the counterfactual by the forest cover loss rate in the treatment area (protected area), or vice-versa when the latter is larger. The same method was used to calculate the treatment effect in terms of carbon emissions. We used the unpaired Wilcoxon test to assess whether there were significant differences between treatment areas and the matched controls.
We used a cleaned map of terrestrial protected areas that were established prior to the year 2000 from the March 2018 World Database on Protected Areas (WDPA; UNEP-WCMC 2018). To identify and remove overlaps between designations (e.g., international and national designation) and IUCN classifications, we created separate layers for each IUCN classification, erasing any overlap between the categories and retaining the strictest classification (Deguignet et al. 2017; see Supplementary Information). The terrestrial administrative boundaries used in the analysis are taken from the Database of Global Administrative Areas (GADM; www.gadm.org), version 2.8, November 2015.
Definition of effectively managed protected areas
The International Union for the Conservation of Nature (IUCN) Green List framework evaluates protected area management effectiveness based on four components: (1) good governance, (2) sound design and planning, (3) effective management, and (4) successful conservation outcomes. It is a new global standard for assessing whether protected areas are achieving conservation outcomes through effective management and equitable governance (IUCN and WCPA 2017). However, because it is new, it has not yet been widely applied in protected area evaluations. The Management Effectiveness Tracking Tool METT;25 is the largest global source of information on protected area management effectiveness63.
We used METT assessments conducted between 2000 and 2014 as protected area management scores. We used the approach outlined in Graham et al.37 to select, exclude, and re-align METT survey responses to the four IUCN Green List Standard components based on congruence between objectives being measured by each indicator (also see Table S11 in Supplementary Information). We also excluded the conservation outcomes survey responses because we replaced this with an independent, quantitative measure of ‘conservation outcomes’, being ‘avoided deforestation’ and ‘avoided carbon emissions’, which is a more powerful indicator than a categorical score. The effective management component had more questions than any other category, therefore we split it into two sub-categories: management resourcing and management processes. We were left with the following four dimensions of management: (1) good governance, (2) sound design and planning, (3) management resourcing, and (4) management processes. Each assessment consists of 30 questions that are scored from 0 (inadequate or non-existing) to 3 (adequate or fully implemented). We selected the earliest possible assessment date because management factors should precede any resulting impact. Finally, we calculated an average score for all METT questions within these four groups.
Statistical matching to assess protected area performance
We selected control sites to measure the impact of protection using the MatchIt package in R64. To account for the non-random distribution of protected areas, we included multiple predictive factors to balance the counterfactual and treatment samples. We based our approach for selecting predictive factors on analyses of deforestation in Southeast Asia24,65 and from global protected area impact studies using matching13,14. We selected anthropogenic drivers (human population density, presence of roads and railways, pressure to expand croplands and urban environments), suitability for land conversion (elevation, slope, and agricultural suitability), and biological aspects (peat characteristics and forest cover; see Table 2 for data sources and Supplementary Information for more details). To account for leakage of protected areas, we excluded buffers of 10 km around their boundaries. The buffer area is inappropriate as an independent control, because the area immediately outside a protection zone can be partially protected due to the close proximity to the restricted area18 or subject to higher deforestation when logging and other activities are displaced to the area immediately outside the protected area66,67. We randomly selected a sample covering 20% of the total study region to reduce spatial autocorrelation68. After removing missing values, we retained a final sample of 747,714 of 1 km2 pixels.
We used Propensity Score Matching (PSM) to create a statistically balanced counterfactual sample to evaluate protected area impact for each country. Despite its limitations, PSM remains the most widely used matching approach14,50,68 and often performs comparatively better than other methods when handling a large numbers of covariates70. We tested coarsened exact matching (CEM) methods, however PSM performed better at finding paired matches for each pixel while maximising sample size. The unprotected pixels included in the matched analysis, represented the pixels most comparable to the protected pixels with respect to the matching variables within the same country (Fig. 6) using the nearest neighbour method, without replacement. We matched each treatment pixel to a unique control observation that had not been matched previously. We calculated avoided forest cover loss and carbon emissions at a pixel-by-pixel level, using one-for-one pairs.
Approach for linking forest conservation to management
The forest cover loss rates during the period 2000–2018 and associated carbon emissions for each protected area was our unbiased measure of avoided forest cover loss and carbon emissions from protected areas. We built two predictive linear models that tested the direction and strength of the relationship between management and contextual factors, and the two dependent variables (1) avoided forest cover loss; and (2) avoided carbon emissions inside protected areas. Southeast Asian deforestation and degradation is largely driven by agriculture and logging, therefore anthropogenic drivers, accessibility and peat characteristics are the best determinants of the spatial patterns of deforestation24. Our independent variables included four management factors: (1) good governance, (2) sound design and planning, (3) management resourcing, and (4) management processes, as well as (5) years protected, (6) protected area size, (7) human pressure, (8) elevation, (9) slope, (10) agricultural suitability, (11) national government transparency, and (12) GDP (Supplementary Information). We selected the data that was nearest to the year 2000 for our contextual variables, prioritising older over more recent data, because management and contextual factors should precede any resulting impact. We performed correlation tests on all model variables. This resulted in removing GDP, elevation, and suitability for growing rice, cassava and maize due to high intercorreality. Slope and size were log-transformed to the power of two, to prevent outliers driving the results. The best-fit model was determined based on Akaike Information Criterion (AIC) of all possible configurations of predictor variables, using the MuMIn package71.
The datasets used in this study are publicly available and may be provided by the corresponding author on reasonable request.
Gibson, L. et al. Primary forests are irreplaceable for sustaining tropical biodiversity. Nature 478, 378–383 (2011).
Luyssaert, S. et al. Old-growth forests as global carbon sinks. Nature 455, 213–215 (2008).
WWF. Living planet report 2020 - bending the curve of biodiversity loss. (WWF, Gland, Switzerland, 2020).
Grantham, H. S. et al. Anthropogenic modification of forests means only 40% of remaining forests have high ecosystem integrity. Nat. Commun. 11, 5978. https://doi.org/10.1038/s41467-020-19493-3 (2020).
Steffen, W. et al. Planetary boundaries: Guiding human development on a changing planet. Science 347, 1259855. https://doi.org/10.1126/science.1259855 (2015).
Balmford, A. et al. Economic reasons for conserving wild nature. Science 297, 950–953 (2002).
Hockings, M. Systems for assessing the effectiveness of management in protected areas. Bioscience 53, 823–832. https://doi.org/10.1641/0006-3568(2003)053[0823:Sfateo]2.0.Co;2 (2003).
Reboredo Segovia, A. L., Romano, D. & Armsworth, P. R. Who studies where? Boosting tropical conservation research where it is most needed. Front. Ecol. Environ. 18, 159–166. https://doi.org/10.1002/fee.2146 (2020).
Geldmann, J. et al. Effectiveness of terrestrial protected areas in reducing habitat loss and population declines. Biol. Conserv. 161, 230–238. https://doi.org/10.1016/j.biocon.2013.02.018 (2013).
Heino, M. et al. Forest loss in protected areas and intact forest landscapes: A global analysis. PLoS ONE 10, e0138918. https://doi.org/10.1371/journal.pone.0138918 (2015).
Joppa, L. N. & Pfaff, A. High and far: Biases in the location of protected areas. PLoS ONE 4, e8273. https://doi.org/10.1371/journal.pone.0008273 (2009).
Ferraro, P. et al. More strictly protected areas are not necessarily more protective: Evidence from bolivia, costa rica, indonesia, and thailand. Environ. Res. Lett. 8, 025011 (2013).
Joppa, L. N. & Pfaff, A. Global protected area impacts. Proc. R. Soc. London B Biol. Sci. 278, 1633–1638 (2011).
Geldmann, J., Manica, A., Burgess, N. D., Coad, L. & Balmford, A. A global-level assessment of the effectiveness of protected areas at resisting anthropogenic pressures. Proc. Natl. Acad. Sci. 116, 23209–23215. https://doi.org/10.1073/pnas.1908221116 (2019).
Allan, J. R. et al. Recent increases in human pressure and forest loss threaten many natural world heritage sites. Biol. Conserv. 206, 47–55. https://doi.org/10.1016/j.biocon.2016.12.011 (2017).
Watson, J., Edward, M. & Venter, O. Mapping the continuum of humanity’s footprint on land. One Earth 1, 175–180. https://doi.org/10.1016/j.oneear.2019.09.004 (2019).
Joppa, L. & Pfaff, A. Reassessing the forest impacts of protection. Ann. N. Y. Acad. Sci. 1185, 135–149. https://doi.org/10.1111/j.1749-6632.2009.05162.x (2010).
Gaveau, D. L. A. et al. Evaluating whether protected areas reduce tropical deforestation in sumatra. J. Biogeogr. 36, 2165–2175. https://doi.org/10.1111/j.1365-2699.2009.02147.x (2009).
Andam, K. S., Ferraro, P. J., Pfaff, A., Sanchez-Azofeifa, G. A. & Robalino, J. A. Measuring the effectiveness of protected area networks in reducing deforestation. Proc. Natl. Acad. Sci. 105, 16089–16094. https://doi.org/10.1073/pnas.0800437105 (2008).
Potapov, P. et al. The last frontiers of wilderness: Tracking loss of intact forest landscapes from 2000 to 2013. Sci. Adv. https://doi.org/10.1126/sciadv.1600821 (2017).
Achard, F. et al. Determination of tropical deforestation rates and related carbon losses from 1990 to 2010. Glob. Change Biol. 20, 2540–2554. https://doi.org/10.1111/gcb.12605 (2014).
Hughes, A. C. Understanding the drivers of southeast asian biodiversity loss. Ecosphere 8, e01624. https://doi.org/10.1002/ecs2.1624 (2017).
Sodhi, N. S., Koh, L. P., Brook, B. W. & Ng, P. K. L. Southeast asian biodiversity: An impending disaster. Trends Ecol. Evol. 19, 654–660. https://doi.org/10.1016/j.tree.2004.09.006 (2004).
Estoque, R. C. et al. The future of southeast asia’s forests. Nat. Commun. 10, 1829–1829. https://doi.org/10.1038/s41467-019-09646-4 (2019).
Stolton, S. et al. Reporting Progress in Protected Areas a Site Level Management Effectiveness Tracking Tool (Gland, 2007).
Coad, L. et al. Measuring impact of protected area management interventions: Current and future use of the global database of protected area management effectiveness. Philos. Trans. R. Soc. B Biol. Sci. 370, 20140281. https://doi.org/10.1098/rstb.2014.0281 (2015).
CBD. Cop 10 decision x/2: Strategic Plan for Biodiversity 2011–2020 (Convention on Biological Diversity, 2011).
UNFCCC. Adoption of the Paris Agreement (Proposal by the President Draft Decision -/CP.21, 2015).
Gaveau, D. L. A. et al. Four Decades of Forest Persistence, Clearance and Logging on Borneo. Vol. 9 (2014).
Bebber, D. P. & Butt, N. Tropical protected areas reduced deforestation carbon emissions by one third from 2000–2012. Sci. Rep. 7, 14005. https://doi.org/10.1038/s41598-017-14467-w (2017).
Buřivalová, Z., Hart, S. J., Radeloff, V. C. & Srinivasan, U. Early warning sign of forest loss in protected areas. Curr. Biol. https://doi.org/10.1016/j.cub.2021.07.072 (2021).
Apan, A., Suarez, L. A., Maraseni, T. & Castillo, J. A. The rate, extent and spatial predictors of forest loss (2000–2012) in the terrestrial protected areas of the philippines. Appl. Geogr. 81, 32–42. https://doi.org/10.1016/j.apgeog.2017.02.007 (2017).
Graham, V., Nurhidayah, L. & Astuti, R. Reference Module in Earth Systems and Environmental Sciences (Elsevier, 2019).
Graham, V., Laurance, S. G., Grech, A., McGregor, A. & Venter, O. A comparative assessment of the financial costs and carbon benefits of redd+ strategies in southeast asia. Environ. Res. Lett. 11, 114022. https://doi.org/10.1088/1748-9326/11/11/114022 (2016).
Mascia, M. B. et al. Protected area downgrading, downsizing, and degazettement (paddd) in africa, asia, and latin america and the caribbean, 1900–2010. Biol. Conserv. 169, 355–361. https://doi.org/10.1016/j.biocon.2013.11.021 (2014).
Geldmann, J. et al. A global analysis of management capacity and ecological outcomes in terrestrial protected areas. Conserv Lett 11, e12434 (2018).
Graham, V. et al. Management resourcing and government transparency are key drivers of biodiversity outcomes in southeast asian protected areas. Biol. Conserv. 253, 108875. https://doi.org/10.1016/j.biocon.2020.108875 (2021).
Gill, D. A. et al. Capacity shortfalls hinder the performance of marine protected areas globally. Nature 543, 665 (2017).
Coad, L. et al. Widespread shortfalls in protected area resourcing undermine efforts to conserve biodiversity. Front Ecol Environ 17, 259–264. https://doi.org/10.1002/fee.2042 (2019).
Carranza, T., Manica, A., Kapos, V. & Balmford, A. Mismatches between conservation outcomes and management evaluation in protected areas: A case study in the brazilian cerrado. Biol. Conserv. 173, 10–16. https://doi.org/10.1016/j.biocon.2014.03.004 (2014).
Nolte, C. & Agrawal, A. Linking management effectiveness indicators to observed effects of protected areas on fire occurrence in the amazon rainforest. Conserv. Biol. 27, 155–165. https://doi.org/10.1111/j.1523-1739.2012.01930.x (2013).
Nolte, C., Agrawal, A. & Barreto, P. Setting priorities to avoid deforestation in amazon protected areas: Are we choosing the right indicators?. Environ. Res. Lett. 8, 015039. https://doi.org/10.1088/1748-9326/8/1/015039 (2013).
Eklund, J., Coad, L., Geldmann, J. & Cabeza, M. What constitutes a useful measure of protected area effectiveness? A case study of management inputs and protected area impacts in madagascar. Conserv. Sci. Pract. https://doi.org/10.1111/csp2.107 (2019).
Bennett, N. J. et al. Conservation social science: Understanding and integrating human dimensions to improve conservation. Biol. Conserv. 205, 93–108. https://doi.org/10.1016/j.biocon.2016.10.006 (2017).
Schleicher, J., Peres, C. A. & Leader-Williams, N. Conservation performance of tropical protected areas: How important is management?. Conserv. Lett. https://doi.org/10.1111/conl.12650 (2019).
Baccini, A. et al. Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nat. Clim. Change 2, 182–185 (2012).
Walker, W. S. et al. The role of forest conversion, degradation, and disturbance in the carbon dynamics of amazon indigenous territories and protected areas. Proc. Natl. Acad. Sci. 117, 3015–3025. https://doi.org/10.1073/pnas.1913321117 (2020).
Wolosin, M. & Harris, N. Tropical Forests and Climate Change: The Latest Science (World Resources Institute, 2018).
Griscom, B. W. et al. Natural climate solutions. Proc. Natl. Acad. Sci. 114, 11645–11650. https://doi.org/10.1073/pnas.1710465114 (2017).
Schleicher, J. et al. Statistical matching for conservation science. Conserv. Biol. https://doi.org/10.1111/cobi.13448 (2019).
Rights and Resources Initiative. Who owns the world’s land? A global baseline of formally recognized Indigenous and community land rights. (Rights and Resources Initiative, Washington DC, 2015).
Santika, T. et al. Community forest management in indonesia: Avoided deforestation in the context of anthropogenic and climate complexities. Glob. Environ. Chang. 46, 60–71. https://doi.org/10.1016/j.gloenvcha.2017.08.002 (2017).
Dudley, N., Shadie, P. & Stolton, S. Guidelines for Applying Protected Area Management Categories Including IUCN WCPA Best Practice Guidance on Recognising Protected Areas and Assigning Management Categories and Governance Types. (IUCN, 2013).
Nelson, A. & Chomitz, K. M. Effectiveness of strict vs. Multiple use protected areas in reducing tropical forest fires: A global analysis using matching methods. PLoS ONE 6, e22722, https://doi.org/10.1371/journal.pone.0022722 (2011).
Ferraro, P. J., Hanauer, M. M. & Sims, K. R. E. Conditions associated with protected area success in conservation and poverty reduction. Proc. Natl. Acad. Sci. https://doi.org/10.1073/pnas.1011529108 (2011).
Oldekop, J. A., Holmes, G., Harris, W. E. & Evans, K. L. A global assessment of the social and conservation outcomes of protected areas. Conserv. Biol. 30, 133–141. https://doi.org/10.1111/cobi.12568 (2016).
Buchner, B. et al. The Global Landscape of Climate Finance 2015 (Climate Policy Initiative, 2015).
Climate Focus. Progress on the New York Declaration on Forests: Finance for Forests (Climate Focus, 2017).
Scharlemann, J. P. W. et al. Securing tropical forest carbon: The contribution of protected areas to redd. Oryx 44, 352–357 (2010).
Hansen, M. C. et al. High-resolution global maps of 21st-century forest cover change. Science 342, 850–853. https://doi.org/10.1126/science.1244693 (2013).
Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A. & Hansen, M. C. Classifying drivers of global forest loss. Science 361, 1108–1111. https://doi.org/10.1126/science.aau3445 (2018).
Zarin, D. J. et al. Tree Biomass Loss: CO2 Emissions from Aboveground Woody Biomass Loss in the Tropics. www.globalforestwatch.org (2020).
Coad, L. et al. Measuring impact of protected area management interventions: Current and future use of the global database of protected area management effectiveness. Philos. Trans. R. Soc. London B Biol. Sci. 370 (2015).
Ho, D., Imai, K., King, G. & Stuart, E. Matchit: Nonparametric preprocessing for parametric causal inference. J. Stat. Softw. 42 (2011).
Hosonuma, N. et al. An assessment of deforestation and forest degradation drivers in developing countries. Environ. Res. Lett. 7, 044009. https://doi.org/10.1088/1748-9326/7/4/044009 (2012).
Ewers, R. M. & Rodrigues, A. S. Estimates of reserve effectiveness are confounded by leakage. Trends Ecol. Evol. 23, 113–116 (2008).
Oliveira, P. J. et al. Land-use allocation protects the peruvian amazon. Science 317, 1233–1236 (2007).
Negret, P. J. et al. Effects of spatial autocorrelation and sampling design on estimates of protected area effectiveness. Conserv. Biol. 34, 1452–1462. https://doi.org/10.1111/cobi.13522 (2020).
Miettinen, J., Shi, C., Tan, W. J. & Liew, S. C. 2010 land cover map of insular southeast asia in 250-m spatial resolution. Remote Sens. Lett. 3, 11–20. https://doi.org/10.1080/01431161.2010.526971 (2012).
Stuart, E., Rubin, D. & Osborne, J. Best Practices in Quantitative Methods (Sage Publications, 2007).
Barton, K. & Barton, M. K. Package ‘mumin’. Version 1, 18 (2015).
We give special thanks to everyone who collected data and made it freely available to use. We also specifically acknowledge and thank the following people. Stuart Allen from Macquarie University for technical support preparing the samples. Marine Deguignet from UNEP-WCMC for help with accessing the METT and WDPA datasets. Zin Phyo Han Tun from the Ministry of Natural Resources and Environmental Conservation, Republic of the Union of Myanmar for sharing data. Marc Hockings and Sue Stolton from the IUCN World Commission on Protected Areas for valuable comments on re-aligning the management effectiveness data. The authors were supported by an Australian Government RTP Scholarship and the Marie Skłodowska-Curie Program (No 676108), Villum Fonden (VKR023371).
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Graham, V., Geldmann, J., Adams, V.M. et al. Southeast Asian protected areas are effective in conserving forest cover and forest carbon stocks compared to unprotected areas. Sci Rep 11, 23760 (2021). https://doi.org/10.1038/s41598-021-03188-w
This article is cited by
Nature Communications (2023)
High deforestation trajectories in Cambodia slowly transformed through economic land concession restrictions and strategic execution of REDD+ protected areas
Scientific Reports (2022)
Nature Communications (2022)